Shein’s massive scale and Darwinian feedback loops

Shein – the Chinese fast fashion company – has six hundred thousand items in it’s inventory at a time and adds another 6000 SKUs every day. That’s mad scale. Imagine having to manage that stock, that inventory, buying ordering, returns, quality control, invoices, everything. You have let go of the idea of managing it all and let computers do what computers do well. Manage the computers, the computers manage the processes.

Letting go of the ideas that stop you taking full advantage of digital. Thats the hard bit. And doing it will challenge the essence of what you do.

Realtime integrated feedback

Rest of World describes Shein’s “large-scale automated test and re-order (LATR) model.”:

The company confirmed it starts by ordering a small batch of each garment, often a few dozen pieces, and then waits to see how buyers respond. If the cropped sweater vest is a hit, Shein orders more. It calls the system a “large-scale automated test and re-order (LATR) model.”

And those factories are fully integrated:

There are 1000s of ‘ghost factories’ all running archaic systems. SHEIN promises demand but in exchange installs its order systems at the factory and has complete visibility into their supply chain. SHEIN pays on time, teaches factories how to run efficient real-time manufacturing operations, and in return, has a loyal base of supply that reacts to demand – in real-time.

They install their systems into the factory and have “complete visibility” the factories “react to demand in real time”. So it’s not hard to imagine that if a shirt sells on the app, that data causes more of those shirts to be ordered in realtime.

Feedback loops without people involved – less ‘test and learn’ and more ‘test and react’.

Minimum viable products

“[Shein] starts by ordering a small batch of each garment, often a few dozen pieces, and then waits to see how buyers respond.” Each product is brought to market as cheaply as possible and as a test. Why do more? At this point it’d feel irresponsible to do more. The cost of failing will be tiny, and it’d be easy to imagine a system where the insight created from failure is worth more than the cost of the experiment.

All that’s left is generate lots of designs for garments. Get the them on the app and let the fittest survive.